Estimands can be used in studies of healthcare interventions to clarify the interpretation of treatment effects. The addendum to the ICH E9 harmonised guideline on statistical principles for clinical trials (ICH E9(R1)) describes a framework for using estimands as part of a study. This paper provides an overview of the estimands framework, as outlined in the addendum, with the aim of explaining why estimands are beneficial; clarifying the terminology being used; and providing practical guidance on using estimands to decide the appropriate study design, data collection, and estimation methods. This article illustrates how to use the estimands framework by applying it to an ongoing trial in emergency bowel surgery. Estimands can be a useful way of clarifying the exact research question being evaluated in a study, both to avoid misinterpretation and to ensure that study methods are aligned to the overall study objectives.
Research studies are often used to answer questions about the effect of an intervention. However, deciding on the precise research question to ask, or how best to design the study to answer that question, can be challenging. Consider the FLO-ELA trial, a pragmatic trial comparing two methods of fluid delivery (cardiac output monitor v clinician judgment) in patients undergoing emergency bowel surgery.1 Because of the lead-in time required to prepare the intervention, a small delay between randomisation and the start of surgery is necessary, and so some participants in FLO-ELA could have their surgery cancelled after randomisation, either because they become too unwell or the underlying issue has resolved itself.
A standard approach for pragmatic trials is to conduct an intention-to-treat analysis, which would include participants who did not undergo surgery.2 However, consideration of the research question could lead investigators to question this approach. An intention-to-treat analysis answers the question “What is the difference between …